CN103927784A - Three-dimensional scanning method - Google Patents
Three-dimensional scanning method Download PDFInfo
- Publication number
- CN103927784A CN103927784A CN201410155804.9A CN201410155804A CN103927784A CN 103927784 A CN103927784 A CN 103927784A CN 201410155804 A CN201410155804 A CN 201410155804A CN 103927784 A CN103927784 A CN 103927784A
- Authority
- CN
- China
- Prior art keywords
- data
- motion
- scanning method
- scanning
- continuous
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 82
- 230000033001 locomotion Effects 0.000 claims abstract description 80
- 230000003993 interaction Effects 0.000 claims abstract description 28
- 238000004458 analytical method Methods 0.000 claims abstract description 9
- 230000004927 fusion Effects 0.000 claims abstract description 5
- 230000008569 process Effects 0.000 claims description 30
- 230000003068 static effect Effects 0.000 claims description 25
- 239000012634 fragment Substances 0.000 claims description 16
- 230000008859 change Effects 0.000 claims description 13
- 238000006243 chemical reaction Methods 0.000 claims description 12
- 238000000059 patterning Methods 0.000 claims description 10
- 230000001154 acute effect Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 7
- 230000007704 transition Effects 0.000 claims description 7
- 238000013480 data collection Methods 0.000 abstract description 2
- 238000005192 partition Methods 0.000 abstract 1
- 230000008846 dynamic interplay Effects 0.000 description 8
- 230000000007 visual effect Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 241001269238 Data Species 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000000903 blocking effect Effects 0.000 description 2
- 230000001186 cumulative effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000011387 Li's method Methods 0.000 description 1
- 241000258241 Mantis Species 0.000 description 1
- 206010028980 Neoplasm Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 229940064452 artec Drugs 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000002459 sustained effect Effects 0.000 description 1
Landscapes
- Processing Or Creating Images (AREA)
Abstract
The invention provides a three-dimensional scanning method. The three-dimensional scanning method comprises the following steps that S101, data collection is conducted, wherein continuous scanning is conducted to obtain continuous data of an object, and the continuous data comprise outer geometric surface data obtained when the object is stationary, movement scanning data obtained when a user interacts with the object or a scene, and hidden data of the interior and a shielded portion of the object; S103, data segment partition is conducted, wherein the continuous data are partitioned into segments with user interaction and segments without user interaction; S105, interaction movement analysis is conducted, wherein user interaction is detected in all segments with user interaction, and corresponding movement scanning data are removed; S107, data fusion is conducted, wherein registration is conducted on all the outer geometric surface data obtained when the object is stationary, and all the hidden data of the interior and the shielded portion of the object, and a three-dimensional model of the object is reconstructed. The three-dimensional scanning method can effectively solve the problem that data of the three-dimensional object are difficult to completely capture because the interior structure and other invisible portions of the object exist.
Description
Technical field
The present invention relates to computer image technology field, relate in particular to a kind of 3-D scanning method.
Background technology
Carry out the measurement of object and the reconstruction background that has a wide range of applications based on 3 d scan data, be subject to giving more sustained attention of academic research and commercial development always.Along with the fast development of 3-D scanning technology in recent years, the scanner of various precision, flexibility ratio and price is come out one after another.Wherein be useful on the vehicle-mounted or airborne scanner that large scale scene is caught, have the high speed scanner of fast Acquisition motion, have the scanner of high precision catches surface details.Space data collection technology also will be in resolution, speed, precision constantly progressive and development, and then open a new situation for the application of 3-D technology.
Stationary body is operated in development both domestic and external and comprises with complete obtaining that dynamically can changing object:
Stationary body scanning
Due to scanner at every turn can only scanning object a side, so if complete scan object must carry out various visual angles scanning to it, then the scan-data at each visual angle is registrated to the scan-data that forms together this complete object.The registration of various visual angles data is mainly used iterative closest point algorithms ICP(Iterative Closest Point), this algorithm is the earliest by Besl and McKay(BESL, P.J., AND MCKAY, N.D.1992.A method for registration of3-D shapes.IEEE Trans.PAMI14, 2, 239 – 256.) and Chen and Medioni(CHEN, Y., AND MEDIONI, G.1992.Object modelling by registration of multiple range images.Image and Vision Computing10, 3, 145 – 155.) propose, for having the scan-data of enough laps or grid to align by two.In the case of the position of the general alignment of known two data, ICP is through finding iteratively matching double points, and calculates according to this rigid transformation of two data, finally converges to the position of alignment.In order to rebuild complete object model, Chen and Medioni little by little snap to new scan-data in all data before.
Even if but every two data can snap to ideal position, because the reason of noise and shortage of data, also can produce small error, the accumulation of these errors will cause the problem of circuit error (Loop Closure), be that top data and last data should overlap, but finally cannot align due to cumulative errors.For this reason, the people such as Bergevin (BERGEVIN, R., S OUCY, M., G AGNON, H., AND LAURENDEAU, D.1996.Towards a general multi-view registration technique.IEEE Trans.PAMI18, 5, 540 – 547.), the people such as Pulli (PULLI, K.1999.Multiview registration for large data sets.In Proceedings of the2nd international conference on3-D digital imaging and modeling.) and the people (BROWN such as Brown, B.J., AND RUSINKIEWICZ, S.2007:Global non-rigid alignment of3-dscans.In ACM SIGGRAPH2007Papers, ACM, New York, NY, USA, SIGGRAPH ' 07.) by carrying out registration by logical each scan-data with its overlapping every other data, thereby cumulative errors is evenly diffused in each data, obtain correct result with this.
People (the WEISE such as Weise afterwards, T., WISMER, T., LEIBE, B., ANDGOOL, L.V.2011.Online loop closure for real-time interactive3d scanning.Comput.Vis.Image Underst.115,5 (May)) develop the method that solves real time scan system Middle Ring Line mis-tie problem.The nonlinear characteristic of scanning device and calibrated error can cause the low-frequency distortion of object scan-data conventionally simultaneously.In order to correct this distortion, the people such as Ikemoto (IKEMOTO, L., G ELFAND, N., AND LEVOY, M.2003.A hierarchical method for aligning warped meshes.In Proc.3DIM.) have introduced non-linear registration technology.This technology is divided scan-data to make multiple overlapping fritters, and these fritters are carried out to overall rigidity alignment.Although do not require specific distorted characteristic, but this method can cause net result too smooth, and be square level of fritter number working time.The people such as Brown utilize thin plate spline to represent this low-frequency distortion, first obtain the sparse correspondence between each visual angle scan-data by the ICP method of partial weight, and calculate the final position of unique point, finally by the method for thin model's bar, all somes data are mapped on its final position.
Dynamically can changing object scanning
For the object that there will be self change of shape in scanning process, we are unified is classified as this type of.Method is to utilize a priori for object self intuitively, and such as a general template of object, or supposition object is that radial type structure (for example human body) is simplified this complex nature of the problem.Pekelny and Gostsman(PEKELNY, Y., AND GOTSMAN, C.2008.Articulated object reconstruction and markerless motion capture from depth video.Computer Graphics Forum27,2 (Apr.), 399 – 408.) propose for the motion tracking of radial type structure object and the method for registration, skeleton and each rigid element of the method hypothesis object are given.Similarly, Chang and Zwicker(CHANG, W., AND ZWICKER, M.2011.Global registration of dynamic range scans for articulated model reconstruction.ACM Trans.Graph.30,3 (May), 26:1 – 26:15.) convert the registration of the scan-data that carries out radial type object by finding the sectional rigid of one group of optimum.The people such as Chang utilizes again deformable template to solve the registration problems with the radial type object of shortage of data afterwards, by the parameter of optimizing deformable template, all scan-datas is registrated to together.The people such as Li (LI, H., LUO, L., VLASIC, D., PEERS, P., POPOVI, J., PAULY, M., AND RUSINKIEWICZ, S.2012.Temporally coherent completion of dynamic shapes.ACM Trans.Graph.31,1 (Feb.), 2:1 – 2:11.) utilize coarse template for instruct carry out can changing object the reconstruction of motion, finally utilize scan-data to obtain comparatively accurate model detail.
Can obtain good effect although utilize the priori of object to carry out problem reduction, these methods but cannot be general, so some algorithms that directly utilize scan-data to carry out deformable object volume reconstruction are proposed in succession.The people such as Mitra (MITRA, N.J., F LORY, S., OVSJANIKOV, M., G ELFAND, N., GUIBAS, L., AND POTTMANN, H.2007.Dynamic geometry registration.In Proceedings of the fifth Eurographics symposium on Geometry processing, Eurographics Association, Airela-Ville, Switzerland, Switzerland, SGP ' 07,173 – 182.) utilize the kinetic property of four dimensional spacetime surface data follow the tracks of can changing object with registration multiframe data.Similarly, the people such as Wand (WAND, M., J ENKE, P., H UANG, Q., B OKELOH, M., G UIBAS, L., AND SCHILLING, A.2007.Reconstruction of deforming geometry from time-varying point clouds.In Proceedings of the fifth Eurographics symposium on Geometry processing, Eurographics Association, Aire-la-Ville, Switzerland, Switzerland, SGP ' 07, 49 – 58.) rebuild scan-data by four-dimensional shape optimized algorithm shape with and movement locus, after work in, they are defined as apparent motion displacement field and calculate the model (WAND that can mate with scan-data, M., A DAMS, B., OVSJANIKOV, M., B ERNER, A., BOKELOH, M., J ENKE, P., G UIBAS, L., S EIDEL, H.-P., AND SCHILLING, A.2009.Efficient reconstruction of nonrigid shape and motion from real-time3d scanner data.ACM Trans.Graph.28, 2 (May), 15:1 – 15:15.).The people such as Sumuth (SUMUTH, J., W INTER, M., AND GREINER, G.2008.Reconstructing animated meshes from time-varying point clouds.In Proceedings of the Symposium on Geometry Processing, Eurographics Association, Aire-la-Ville, Switzerland, Switzerland, SGP ' 08, 1469 – 1476.) and the people (SHARF such as Sharf, A., A LCANTARA, D.A., L EWINER, T., G REIF, C., SHEFFER, A., A MENTA, N., AND COHEN-O R, D.2008.Space-time surface reconstruction using incompressible flow.In ACM SIGGRAPH Asia2008papers, ACM, New York, NY, USA, SIGGRAPH Asia ' 08, 110:1 – 110:10.) directly recover four-dimensional surface model by implicit expression spatiotemporal motion expression way.But these technology are not only consuming time but also need to differ between adjacent data can not be too large.
The people such as Li (LI, H., S UMNER, R.W., AND PAULY, M.2008.Global correspondence optimization for non-rigid registration of depth scans.In Proceedings of the Symposium on Geometry Processing, Eurographics Association, Aire-la-Ville, Switzerland, Switzerland, SGP ' 08, 1421 – 1430.) one non-rigid registration method between two proposed in 2007, non-rigid deformation is described as the deformation problems based on figure by the method, by global optimization, obtain lap and the alignment conversion of two data simultaneously.This algorithm is for the little relatively robust of two data of deformation.Afterwards, in the application that the three-dimensional that this algorithm is expanded to completion, the personage of dynamic body different conditions data within continuous time by the people such as Li is autodyned and the analysis of cancer detection data etc. is of practical significance.Similar with the people's such as Li method, the people such as Huang (HUANG, Q.-X., A DAMS, B., W ICKE, M., AND GUIBAS, L.J.2008.Non-rigid registration under isometric deformations.In Proceedings of the Symposium on Geometry Processing, Eurographics Association, Aire-la-Ville, Switzerland, Switzerland, SGP ' 08, 1449 – 1457.) suppose that the shape of object becomes equidistant variation, and with the geodesic distance of the formal description object data of figure, and then extract one group of two geodetic corresponding point between data, utilize these corresponding point to carry out registration to data.These between two non-rigid registration method may be used to rebuild motion that can changing object.
But in the scan method of traditional three-dimensional body, no matter be stationary body or dynamic object, current all technology can only be used for the complete of the external surface of object and obtain.But in many application, the complete information of object not only comprises external surface, more comprise internal structure of body or because other block former thereby to being scanned the sightless part of instrument.For example rebuild an Office Building, except the metope of outside, building surrounding, the more important thing is office, the hall of floor inside; For another example a cabinet is except door, handle and other peripheries of outside, and it is also extremely important that the dividing plate of cabinet inside or inside are the article of putting; But more complicated very important example is houseplant with luxuriant foliage, and its inner branches and leaves all or are partly blocked by outside branches and leaves, and cannot be scanned instrument Direct Acquisition.By method in the past, must first carry out respectively independent scanning to the internal and external parts of these objects, then inside and outside scan-data manual registration is arrived together.But conventionally because not having lap, inside and outside data make quite difficulty and cannot ensure the correctness of result of registration work.For the situation of plant, it is even worse that problem becomes.Still there is bottleneck in the whole partial datas that therefore, how intactly to obtain three dimensional object.
Known, in existing scan method, there is following defect:
Quality of scanning is low: scanning process covers whole sweep object, and article for rotation or scanner are carefully sampled to object all surface from multiple different visual angles.And in real world, there is hidden part in complex-shaped object, as physics blocks etc., especially in three dimensional object self structure, there is large-area concave surface, hidden inner structure, cannot be obtained the fold of data and be blocked etc. by scanner, makes inevitably to exist in sampled data cavity and the too sparse problem of sampling, thereby affect quality of scanning, greatly reduce the degree of accuracy of data.
Scanning process is numerous and diverse: whole scanning process is very heavy complexity, on the one hand, the professional knowledge that user should have to a certain degree instructs scanning process, on the other hand, user need to constantly be scanned, check and rescan those incorrect parts, complete tediously long operating process with this.
Summary of the invention
For the problems referred to above, the present invention will be by proposing a kind of 3-D scanning method, there is the problem such as block, shortage of data, precision that the invisible content such as fold, cavity causes are low to solve in prior art object in 3-D scanning process, can be mutual with scene dynamics in scanning process, object or scene internal and external parts are carried out to real continuous complete scan, and catch the data of invisible part, reconstruct the complete three-dimensional model in object inside and outside.
A kind of 3-D scanning method, it comprises the steps:
S101, data acquisition, continuous sweep obtains the continuous data of object, and described continuous data comprises that outer geometric jacquard patterning unit surface data, user and object when object is static or scene carry out motion scan-data when mutual and the hiding data of interior of articles and the portion that is blocked;
S103, data segment are cut apart, and described continuous data is divided into the fragment that there is no user interactions and the fragment that has user interactions;
S105, reciprocal motion analysis detect the mutual of user and reject corresponding motion scan-data in the described fragment that has user interactions;
S107, data fusion, the hiding data of the outer geometric jacquard patterning unit surface data when static to described object and described interior of articles and the portion that is blocked carries out registration, rebuilds the three-dimensional model of object.
In the present invention's one preferred embodiments, in step S101, adopt scanning device to carry out data acquisition to object.
In the present invention's one preferred embodiments, in step S103, whether occur that by detecting continuously motion or acute variation have the fragment of user interactions described in cutting apart in described continuous data.
In the present invention's one preferred embodiments, step S105 further comprises:
S1051, corresponding point are estimated, adopt global rigid conversion and local non-rigid conversion to estimate corresponding point, obtain transition matrix;
S1053, carry out registration according to described transition matrix, calculate new corresponding point, and carry out iteration with this;
S1055, to the continuous place of motion or acute variation in described continuous data, by calculating trajectory clustering, analyze motion, detecting motion and change;
S1057, obtain movement locus, detect dynamic part, reject described dynamic part, retain the static part in motion process.
In the present invention's one preferred embodiments, in step S105, detect described motion scan-data by the track that calculates corresponding point motion in described continuous data.
In the present invention's one preferred embodiments, step S107 adopts the non-rigid change algorithm of robust to carry out registration.
The 3-D scanning method that application the present invention proposes, its beneficial effect is embodied in:
The 3-D scanning method that the present invention proposes, the problem that is difficult to the total data that catches three-dimensional body that can effectively solve owing to blocking, the existence of the invisible part such as fold, concave surface, hiding inner structure causes.
The 3-D scanning method that the present invention proposes is one object or scene internal and external parts is carried out to real continuous complete scan stream, adopt and keep scene stillness and the data of motion scan instrument catches external surface and invisible part, keep scanner static permission user and scene to operate alternately the mode of capture movement process, by this bridge of dynamic interaction, connect and there is no overlapping inside and outside two parts (or outside surface and invisible part), realization truly the obtaining of partial data of object or scene.
In addition, the 3-D scanning method that the present invention proposes, processing in the motion analysis and rejecting process of dynamic interaction, detects action by calculating movement locus, algorithm simple efficient, be easy to realize.Simultaneously, described 3-D scanning method is not only applicable to rigid objects (as cabinet, toilet, automobile trunk, chest etc.), can obtain very complete three-dimensional model for non-rigid feasible change object (plant of blocking as blade, tablecloth, curtain etc.) yet.
Above-mentioned explanation is only the general introduction of technical solution of the present invention, in order to better understand technological means of the present invention, and can be implemented according to the content of instructions, and for above and other objects of the present invention, feature and advantage can be become apparent, below especially exemplified by embodiment, and coordinate accompanying drawing, be described in detail as follows.
Brief description of the drawings
Fig. 1 is the process flow diagram of 3-D scanning method provided by the invention;
Fig. 2 is the process flow diagram that shown in Fig. 1, in 3-D scanning method, reciprocal motion is analyzed;
Fig. 3 is the analytic process schematic diagram that shown in Fig. 1, in 3-D scanning method, reciprocal motion is analyzed;
Fig. 4 is corresponding point estimation and registration schematic diagram in 3-D scanning method shown in Fig. 1;
Fig. 5 is movement locus schematic diagram in 3-D scanning method shown in Fig. 1;
Fig. 6 is the movement locus schematic diagram of cabinet;
Fig. 7 is data acquisition and the autoregistration result figure of cabinet;
Fig. 8 is that 3-D scanning method shown in application drawing 1 is carried out the illustraton of model of rebuilding after 3-D scanning to cabinet;
Fig. 9 is that 3-D scanning method shown in application drawing 1 is carried out the illustraton of model of rebuilding after 3-D scanning to plant;
Figure 10 is that 3-D scanning method shown in application drawing 1 is carried out the illustraton of model of rebuilding after 3-D scanning to tablecloth.
Embodiment
Below in conjunction with drawings and the specific embodiments, the present invention is further detailed explanation.
Refer to Fig. 1, the invention provides a kind of 3-D scanning method, it comprises the steps:
S101, data acquisition, continuous sweep obtains the continuous data of object, and described continuous data comprises that outer geometric jacquard patterning unit surface data, user and object when object is static or scene carry out motion scan-data when mutual and the hiding data of interior of articles and the portion that is blocked.
In the present embodiment, adopt active scan mode, user's (being scanning device operator) at any time gated sweep equipment position and towards, so that object (and/or scene) is scanned flexibly, gathered described continuous data.Be understandable that, all motion scan equipment that can effectively catch motion is all applicable to described data acquisition, consider the precision of current scanning device self and the restriction of visual field, for object and the resolution requirement of different scale, conventionally need to adopt different scanning devices, for the object of different scale, adopt different scanning devices.In the present embodiment, use Mantis F5 hand held scanner to scan medium object, and for more pocket object, adopt Artec Eva hand held scanner.
In the present embodiment, the continuous data that data acquisition obtains comprises that outer geometric jacquard patterning unit surface data acquisition, user and object when object is static or scene carry out motion scan-data when mutual and the hiding data of interior of articles and the portion that is blocked.
Wherein, when the outer geometric jacquard patterning unit surface of the static state of object is scanned, keep object or scene stillness, motion scan instrument, makes all visual angles of its coverture external surface, can obtain outer geometric jacquard patterning unit surface data, for the follow-up registration that carries out.
Described user and object carry out alternately referring to object/scene to operate, and allow to change scene, carry out certain operations, as open cupboard door, mobile shelter, push plant outer blade etc. aside.Undertaken alternately by user and object, inner structure that scanning object shows effectively, the invisible parts such as part are blocked.When scan operation, keep scanner static, and catch the action of dynamic interaction, can obtain user and object or scene and carry out the motion scan-data when mutual.
When the invisible part of object is scanned, stationary state when scene remains dynamic interaction scanning, inside that motion scan instrument scanning object is hidden, the sightless contents such as part are blocked, to obtain partial data, be the hiding data of interior of articles and the portion that is blocked, for the follow-up registration that carries out.
S103, data segment are cut apart, and described continuous data is divided into the fragment that there is no user interactions and the fragment that has user interactions.
In order to obtain the data in static moment of scanned object, interactive action need to be identified and rejected from scan-data sequence, so obtaining after the continuous data of object, first need interaction data to split.
Inventor herein makes discovery from observation: do not carry out when mutual when user and object/scene, the adjacent data collecting, in overlapping position, does not have continuous acute variation; And carry out in mutual process user and object/scene, the same parts of object can present continuous motion or violent variation in scan-data, thus, whether the present invention occurs that by detecting in described continuous data continuously motion or acute variation have the fragment of user interactions described in cutting apart.
In the process of user interactions, can cause the variation that object is larger, i.e. motion.The part of these motions does not belong to the object data in static moment, in reciprocal process, some data of user also can be gathered together in addition, these are not the data that finally need, but auxiliary we blocked accurately and the middle bridge of the object data registration that is blocked.So first need the interactive portion between user and scanning object to find, be the fragment (object is for static) that there is no user interactions and the fragment that has user interactions by continuous data sequences segmentation.Particularly, the fragment that occurs motion continuously or acute variation in described continuous data is split, as the described fragment that has user interactions, for follow-up reciprocal motion analysis.
Mutual in order to reject, first must in continuous sweep data sequence, the mutual initial sum of consumer positioning stop.Because may existing many places, object blocks, so in a continuous sweep, may occur repeatedly alternately.User interactions can cause data some part suddenly or larger variation, so in this part, we be intended to feature by analyzing interaction data and and nonreciprocal data between difference, detect and the data sequence frame of intercorrelation.
S105, reciprocal motion analysis detect the mutual of user and reject corresponding motion scan-data in the described fragment that has user interactions.
In the present embodiment, in reciprocal motion is analyzed, detect dynamic part by the track (trajectory) that calculates corresponding point motion in scan-data sequence, detect described motion scan-data by the track that calculates corresponding point motion in described continuous data.
The detection of interaction data is core work of the present invention.The present invention is registrated to together more accurate Frame, and the position corresponding relation and the difference that then detect between consecutive frame data detect, and then by analyzing in whole interaction sequence data motion or acute variation continuously.For more accurate carrying out detects and cut apart alternately, need to carry out accurate registration to interaction data sequence.But because object constantly changes in this process, cause simple rigidity and non-rigid registration method to lose efficacy.
Based on this, the present invention carries out the analysis of reciprocal motion by following part, as shown in Figures 2 and 3, comprising:
S1051, corresponding point are estimated, adopt global rigid conversion and local non-rigid conversion to estimate corresponding point, obtain transition matrix.
S1053, carry out registration according to described transition matrix, calculate new corresponding point, and carry out iteration with this.
Corresponding point are estimated to be specially with registration:
For given any two consecutive frame f
k, f
k+1, when initial, with Data Segmentation be that piece is similar, p ∈ f
kcorresponding point q ∈ f
k+1that p is at f
k+1the point that middle distance is nearest, normal direction is similar.
The present invention is in data acquisition, and while carrying out dynamic interaction, existing dynamic part has again static part, therefore adopts global rigid conversion D
rwith local non-rigid conversion D
eestimate corresponding point.First rigidity ICP algorithm carries out global rigid conversion to adjacent two frame data; Due to dynamic interaction, local message converts, so use the local non-rigid conversion based on piece density.In the conversion of block-based local non-rigid, for every a pair of corresponding blocks: calculate block-based rigid transformation
for each piece: calculate
carry out registration according to the transition matrix obtaining, calculate new corresponding point, with this continuous iteration, as shown in Figure 4.
S1055, to the continuous place of motion or acute variation in described continuous data, by calculating trajectory clustering, analyze motion, detecting motion and change.
S1057, obtain movement locus, detect dynamic part, reject described dynamic part, retain the static part in motion process.
Trajectory clustering is removed and is specially with motion:
For continuous motion or change violent place, by calculating trajectory clustering, analyze motion, detect motion and change.Two consecutive frames of given registration, obtain block-based corresponding point set.Along the direction of the polygon vector of frame sequence 3D+ time, connect certain a bit all corresponding point on each successive frame, obtain a movement locus.For a bit, on front n+1 successive frame, the i article of track of corresponding point composition definition of this point
as shown in Figure 5.
Due to the transmission of shaking in motion process, the track obtaining like this has noise, is similar to discrete Laplce, smooth track by space-time Laplace operator.
Obtain movement locus, detect dynamic part, reject these dynamic parts, retain the static part in motion process.Particularly, after obtaining reciprocal motion track, delimit dynamic interaction scope by S1055 step, the data of rejecting within the scope of this are rejected to realize motion.
In the present embodiment, the process of user interactions is to be also blocked object gradually to the visible process of scanner, in this process, the dynamic data that the motion moment scans does not belong to the object data in static moment, and the static data now scanning comprises the data of a part of outside shield portions and the data of a part of part that is blocked before, these static datas have formed the bridge of registration between external data and internal data.Catch owing to also can being scanned equipment with a part for object contact in addition, these data are also moved, and also will be rejected in the lump.
S107, data fusion, the hiding data of the outer geometric jacquard patterning unit surface data when static to described object and described interior of articles and the portion that is blocked carries out registration, rebuilds the three-dimensional model of object.
Data after rejecting are alternately mainly the static datas that comprises object, wherein have the deformation of body and the scanner that in reciprocal process, cause to move the rigid transformation bringing.In the present embodiment, adopt the non-rigid change algorithm of robust that all scan-datas are registrated to together, obtain complete scan data.
In the scan-data of having rejected the mutual stage, after dynamic data, need to carry out non-rigid registration to whole scanning sequence frame.As previously mentioned, in these data, comprise the low-frequency distortion causing because of nonlinear problem or the problem of calibrating of scanning device self, and object self some local deformations in reciprocal process.So can not utilize simply the various visual angles Registration of Measuring Data method of rigidity to process, must utilize non-rigid registration method to carry out registration to obtain the complete of object and scanning accurately to these data.
Data fusion is that the result of 3-D scanning method of the present invention is shown.Outside surface when data acquisition has obtained object or scene static state, the data of internal layer, the hiding data of outer geometric jacquard patterning unit surface data acquisition when object is static and interior of articles and the portion that is blocked, respectively registration.Again by this bridge of dynamic interaction, and detect Static and dynamic part through motion analysis, outside surface and these two of internal layers are seen to do not have overlapping part to couple together from data acquisition, be accurately registered in together, reconstruct true complete three-dimensional model.As shown in Figure 6, taking cabinet as example.
Refer to Fig. 7 and Fig. 8, for adopting described 3-D scanning method to scan scanning process and the reconstructed results of the rigid objects data taking cabinet as representative; Refer to Fig. 9 and Figure 10, for adopting described 3-D scanning method to scan respectively scanning process and the reconstructed results of the non-rigid feasible change object such as plant, tablecloth.Known, described 3-D scanning method has higher feasibility and applicability widely.
The above, only embodiments of the invention, not the present invention is done to any pro forma restriction, although the present invention discloses as above with embodiment, but not in order to limit the present invention, any those skilled in the art, do not departing within the scope of technical solution of the present invention, when can utilizing the technology contents of above-mentioned announcement to make a little change or being modified to the equivalent embodiment of equivalent variations, in every case be not depart from technical solution of the present invention content, any simple modification of above embodiment being done according to technical spirit of the present invention, equivalent variations and modification, all still belong in the scope of technical solution of the present invention.
Claims (6)
1. a 3-D scanning method, is characterized in that, comprises the steps:
S101, data acquisition, continuous sweep obtains the continuous data of object, and described continuous data comprises that outer geometric jacquard patterning unit surface data, user and object when object is static or scene carry out motion scan-data when mutual and the hiding data of interior of articles and the portion that is blocked;
S103, data segment are cut apart, and described continuous data is divided into the fragment that there is no user interactions and the fragment that has user interactions;
S105, reciprocal motion analysis detect the mutual of user and reject corresponding motion scan-data in the described fragment that has user interactions;
S107, data fusion, the hiding data of the outer geometric jacquard patterning unit surface data when static to described object and described interior of articles and the portion that is blocked carries out registration, rebuilds the three-dimensional model of object.
2. 3-D scanning method as claimed in claim 1, is characterized in that, in step S101, adopts scanning device to carry out data acquisition to object.
3. 3-D scanning method as claimed in claim 1, is characterized in that, in step S103, whether occurs that by detecting continuously motion or acute variation have the fragment of user interactions described in cutting apart in described continuous data.
4. 3-D scanning method as claimed in claim 1, is characterized in that, step S105 further comprises:
S1051, corresponding point are estimated, adopt global rigid conversion and local non-rigid conversion to estimate corresponding point, obtain transition matrix;
S1053, carry out registration according to described transition matrix, calculate new corresponding point, and carry out iteration with this;
S1055, to the continuous place of motion or acute variation in described continuous data, by calculating trajectory clustering, analyze motion, detecting motion and change;
S1057, obtain movement locus, detect dynamic part, reject described dynamic part, retain the static part in motion process.
5. 3-D scanning method as claimed in claim 1, is characterized in that, in step S105, detects described motion scan-data by the track that calculates corresponding point motion in described continuous data.
6. 3-D scanning method as claimed in claim 1, is characterized in that, step S107 adopts the non-rigid change algorithm of robust to carry out registration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410155804.9A CN103927784B (en) | 2014-04-17 | 2014-04-17 | A kind of active 3-D scanning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410155804.9A CN103927784B (en) | 2014-04-17 | 2014-04-17 | A kind of active 3-D scanning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103927784A true CN103927784A (en) | 2014-07-16 |
| CN103927784B CN103927784B (en) | 2017-07-18 |
Family
ID=51145995
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410155804.9A Active CN103927784B (en) | 2014-04-17 | 2014-04-17 | A kind of active 3-D scanning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103927784B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105486228A (en) * | 2015-11-25 | 2016-04-13 | 南京林业大学 | Tree target volume real-time measuring method based on two-dimension laser scanner |
| CN108871297A (en) * | 2018-08-17 | 2018-11-23 | 中铁大桥（南京）桥隧诊治有限公司 | Bridge structure size detecting method and system based on 3-D scanning technology |
| CN109154417A (en) * | 2016-04-29 | 2019-01-04 | 因吉纳瑞股份公司 | The improved hand system of 3-D scanning is carried out with smart phone |
| CN111964604A (en) * | 2020-10-20 | 2020-11-20 | 常州市建筑科学研究院集团股份有限公司 | Plane deformation monitoring and measuring method based on image recognition |
| CN112254663A (en) * | 2020-10-20 | 2021-01-22 | 常州市建筑科学研究院集团股份有限公司 | Plane deformation monitoring and measuring method and system based on image recognition |
| CN116402956A (en) * | 2023-06-02 | 2023-07-07 | 深圳大学 | Intelligent driven three-dimensional object interactive reconstruction method, device, equipment and medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2389500A (en) * | 2002-04-20 | 2003-12-10 | Virtual Mirrors Ltd | Generating 3D body models from scanned data |
| CN1747559A (en) * | 2005-07-29 | 2006-03-15 | 北京大学 | Three-dimensional geometric mode building system and method |
| CN101321269A (en) * | 2007-06-05 | 2008-12-10 | 同济大学 | Passenger flow volume detection method and system based on computer vision |
| CN101669146A (en) * | 2007-04-24 | 2010-03-10 | 索尼电脑娱乐公司 | 3d object scanning using video camera and TV monitor |
| CN102054176A (en) * | 2010-12-23 | 2011-05-11 | 中国科学院自动化研究所 | Method used for establishing semantic scene models for scene images of moving targets by utilizing computer |
| CN103065352A (en) * | 2012-12-20 | 2013-04-24 | 北京农业信息技术研究中心 | Plant three-dimensional reconstruction method based on image and scanning data |
| WO2014014562A1 (en) * | 2012-07-20 | 2014-01-23 | Google Inc. | Determining three-dimensional (3d) object data models based on object movement |
| CN103646253A (en) * | 2013-12-16 | 2014-03-19 | 重庆大学 | Bus passenger flow statistics method based on multi-motion passenger behavior analysis |
-
2014
- 2014-04-17 CN CN201410155804.9A patent/CN103927784B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2389500A (en) * | 2002-04-20 | 2003-12-10 | Virtual Mirrors Ltd | Generating 3D body models from scanned data |
| CN1747559A (en) * | 2005-07-29 | 2006-03-15 | 北京大学 | Three-dimensional geometric mode building system and method |
| CN101669146A (en) * | 2007-04-24 | 2010-03-10 | 索尼电脑娱乐公司 | 3d object scanning using video camera and TV monitor |
| CN101321269A (en) * | 2007-06-05 | 2008-12-10 | 同济大学 | Passenger flow volume detection method and system based on computer vision |
| CN102054176A (en) * | 2010-12-23 | 2011-05-11 | 中国科学院自动化研究所 | Method used for establishing semantic scene models for scene images of moving targets by utilizing computer |
| WO2014014562A1 (en) * | 2012-07-20 | 2014-01-23 | Google Inc. | Determining three-dimensional (3d) object data models based on object movement |
| CN103065352A (en) * | 2012-12-20 | 2013-04-24 | 北京农业信息技术研究中心 | Plant three-dimensional reconstruction method based on image and scanning data |
| CN103646253A (en) * | 2013-12-16 | 2014-03-19 | 重庆大学 | Bus passenger flow statistics method based on multi-motion passenger behavior analysis |
Non-Patent Citations (2)
| Title |
|---|
| HAO LI 等: "3D Self-Portraits", 《ACM TRANSACTIONS ON GRAPHICS》 * |
| 数字所可视计算研究中心: "深圳先进院可视计算研究中心两论文入选SIGGRAPH 2014", 《HTTP://SZS.SIAT.AC.CN/CONTENT/VIEWCONTENT.ACTION?ID=378》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105486228A (en) * | 2015-11-25 | 2016-04-13 | 南京林业大学 | Tree target volume real-time measuring method based on two-dimension laser scanner |
| CN105486228B (en) * | 2015-11-25 | 2018-04-03 | 南京林业大学 | A kind of trees target volume method for real-time measurement based on two dimensional laser scanning instrument |
| CN109154417A (en) * | 2016-04-29 | 2019-01-04 | 因吉纳瑞股份公司 | The improved hand system of 3-D scanning is carried out with smart phone |
| CN109154417B (en) * | 2016-04-29 | 2020-10-20 | 因吉纳瑞股份公司 | Improved handheld system for three-dimensional scanning with smart phone |
| CN108871297A (en) * | 2018-08-17 | 2018-11-23 | 中铁大桥（南京）桥隧诊治有限公司 | Bridge structure size detecting method and system based on 3-D scanning technology |
| CN111964604A (en) * | 2020-10-20 | 2020-11-20 | 常州市建筑科学研究院集团股份有限公司 | Plane deformation monitoring and measuring method based on image recognition |
| CN112254663A (en) * | 2020-10-20 | 2021-01-22 | 常州市建筑科学研究院集团股份有限公司 | Plane deformation monitoring and measuring method and system based on image recognition |
| CN111964604B (en) * | 2020-10-20 | 2021-01-29 | 常州市建筑科学研究院集团股份有限公司 | Plane deformation monitoring and measuring method based on image recognition |
| CN112254663B (en) * | 2020-10-20 | 2022-03-18 | 常州市建筑科学研究院集团股份有限公司 | Plane deformation monitoring and measuring method and system based on image recognition |
| CN116402956A (en) * | 2023-06-02 | 2023-07-07 | 深圳大学 | Intelligent driven three-dimensional object interactive reconstruction method, device, equipment and medium |
| CN116402956B (en) * | 2023-06-02 | 2023-09-22 | 深圳大学 | Intelligent driven three-dimensional object interactive reconstruction method, device, equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103927784B (en) | 2017-07-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wolff et al. | Point cloud noise and outlier removal for image-based 3D reconstruction | |
| CN103927784A (en) | Three-dimensional scanning method | |
| Berger et al. | State of the art in surface reconstruction from point clouds | |
| Vanegas et al. | Building reconstruction using manhattan-world grammars | |
| Chauve et al. | Robust piecewise-planar 3D reconstruction and completion from large-scale unstructured point data | |
| Wang | 3D building modeling using images and LiDAR: A review | |
| Grussenmeyer et al. | Comparison methods of terrestrial laser scanning, photogrammetry and tacheometry data for recording of cultural heritage buildings | |
| CN100559398C (en) | Automatic deepness image registration method | |
| CN108053476B (en) | Human body parameter measuring system and method based on segmented three-dimensional reconstruction | |
| Yang et al. | Ransac matching: Simultaneous registration and segmentation | |
| Starck et al. | Spherical matching for temporal correspondence of non-rigid surfaces | |
| Wang et al. | Window detection from mobile LiDAR data | |
| Livesu et al. | Extracting curve-skeletons from digital shapes using occluding contours | |
| Lhuillier et al. | Manifold surface reconstruction of an environment from sparse structure-from-motion data | |
| Wan et al. | Grammar-based 3D facade segmentation and reconstruction | |
| Toldo et al. | Accurate multiview stereo reconstruction with fast visibility integration and tight disparity bounding | |
| Digne et al. | The level set tree on meshes | |
| Mi et al. | 3D reconstruction based on the depth image: A review | |
| Luo et al. | Sparse RGB-D images create a real thing: A flexible voxel based 3D reconstruction pipeline for single object | |
| Rosinol | Densifying sparse vio: a mesh-based approach using structural regularities | |
| Friedman et al. | Real Time Detection of Repeated Structures in Point Clouds of Urban Scenes. | |
| Previtali et al. | Automatic façade segmentation for thermal retrofit | |
| Klose et al. | Reconstructing Shape and Motion from Asynchronous Cameras. | |
| D'Apuzzo et al. | Least squares matching tracking algorithm for human body modeling | |
| CN216692729U (en) | Indoor surveying and mapping trolley |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20140716 Assignee: Shenzhen Zhangchao Technology Co.,Ltd. Assignor: SHENZHEN INSTITUTES OF ADVANCED TECHNOLOGY CHINESE ACADEMY OF SCIENCES Contract record no.: X2023980043397 Denomination of invention: An Active 3D Scanning Method Granted publication date: 20170718 License type: Common License Record date: 20231013 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20140716 Assignee: Shenzhen achievement transformation Center Co.,Ltd. Assignor: SHENZHEN INSTITUTES OF ADVANCED TECHNOLOGY CHINESE ACADEMY OF SCIENCES Contract record no.: X2023980044084 Denomination of invention: An Active 3D Scanning Method Granted publication date: 20170718 License type: Common License Record date: 20231023 |
|
| EE01 | Entry into force of recordation of patent licensing contract |